. Chapter 8 Building Data and Document-Driven Decision Support Systems INTRODUCTION In recent years, most large companies and many other large organizations have implemented database systems called data warehouses and some have also implemented document management and On-line Analytical Processing (OLAP) systems. Some organizations have implemented Business Intelligence (BI) technologies and some have created Executive Information Systems (EIS). Many managers and information systems specialists are interested in learning more about these relatively new types of data-driven and document-driven Decision Support Systems (DSS). For many years, the prospects and problems of providing managers with real-time management information have been discussed and debated (cf., Dearden, 1966). The debate about costs, advantages, problems, and possibilities must continue. Managers need to retrieve and analyze large structured and unstructured data collections for decision support. The expanded DSS framework categorizes business intelligence systems, data warehouses, EIS, spatial DSS, and OLAP systems as data-driven DSS. In general, a data-driven DSS is an interactive computer-based system that helps a decision maker use a very large database of business data and, in some systems, data about the external environment of a company. For example, a system may have data on both a company’s sales and on its competitors’ sales. Some of the data is very detailed transaction data and some is a summary of transactions. In most implementations of data-driven DSS, users of the system can perform unplanned or ad hoc analyses and requests for data. In a data-driven DSS, managers process data to identify facts and draw conclusions about relationships and trends. Data-driven DSS help managers retrieve, display, and analyze historical data. Document-driven DSS are defined as systems that integrate “a variety of storage and processing technologies to provide complete document retrieval and 124 Decision Support Systems analysis.” The Web now provides access to large document databases, including databases of hypertext documents, images, sounds, and video. Examples of documents that could be accessed by a document-driven DSS are policies and procedures, product specifications, catalogs, news stories, and corporate historical documents, including minutes of meetings, corporate records, and important correspondence. A search engine is a powerful decision-aiding tool associated with a document-driven DSS (cf., Fedorowicz, 1993, pp. 125–136; Swanson and Culnan, 1978; Power, 2001). Knowledge management, Web, and DSS technologies are used to build document-driven DSS. Data and document-driven DSS are often very expensive to develop and implement in organizations. Despite the large resource commitments that are required, many companies have implemented these types of DSS. Technologies are changing and managers and MIS staff will need to make continuing investments in these categories of DSS software. So, it is important that managers understand the various terms and systems that use large databases to support management decision making. This chapter emphasizes: comparing data and document-driven DSS; identifying subcategories of data-driven DSS, comparing structured DSS data and operating data, understanding an interconnected data-driven DSS architecture, implementing data and document- driven DSS, and finding success in building DSS with large structured and unstructured databases. Now, let’s begin our exploration of these two general categories of DSS by discussing the differences and similarities between them. COMPARING DATA AND DOCUMENT-DRIVEN DSS Document-driven DSS is a relatively new category of decision support. There are certainly similarities to the more familiar data-driven DSS, but there are also major differences. Document-driven DSS help managers process “soft” or qualitative information, and data-driven DSS help managers process “hard” or numeric data. Both categories of DSS come in various shapes and sizes. Some systems support senior managers and others support functional decision makers on narrowly defined tasks. The Web has increased the need for, and the possibilities associated with, document-driven DSS. A defining difference between the two categories of DSS is that data-driven DSS help managers analyze, display and manipulate large structured data sets that contain numeric and short character strings while document-driven DSS analyze, display, and manipulate text including logical units of text, called documents (cf., Sullivan, 2001). Another defining difference is the analysis tools used for decision support. Data-driven DSS use quantitative and statistical tools for ordering, summarizing, and evaluating the specific contents of a subject-oriented data warehouse. Document-driven DSS use natural language and statistical tools for extracting, categorizing, indexing, and summarizing subject-oriented document warehouses. What are the similarities? First, both systems use databases with very large collections of information to drive or create decision support capabilities. Second, both types of systems require the definition of metadata and the Data and Document-Driven Decision Support Systems 125 cleaning, extraction, and loading of data into an appropriate data management system using an organizing framework or model. Third, building either type of system involves understanding the decision support and information needs of the targeted users. Also, because user needs are hard to anticipate the tendency is to store large amounts of data or documents that may not be immediately needed. Rapid application development or prototyping is sometimes possible for small scale systems, but a more structured SDLC approach is needed for enterprise-wide data or document- driven DSS. Neither type of system can meet all of the decision support needs of all managers in an organization. The best approach is to try to meet a specific, well-defined need initially and then incrementally expand the structured data or documents that are captured and organized in the foundation data/document management system. DATA-DRIVEN DSS SUBCATEGORIES The broad category of data-driven DSS generally includes tools to help users “drill down” for more detailed information, “drill up” to see a broader, more summarized view, and “slice and dice” to change the data dimensions they are viewing. The results of “drilling” and “slicing and dicing” are presented in tables and charts. There are four main subcategories of data-driven DSS; data warehouses, OLAP systems with multidimensional databases, Executive Information Systems (EIS), and spatial DSS. Data Warehouses A data warehouse is a specific database designed and populated to provide decision support in an organization (cf., Gray and Watson, 1998). It is batch- updated and structured for rapid on-line queries and managerial summaries. Data warehouses contain large amounts of data—500 megabytes and more. According to data warehousing pioneer Bill Inmon (1995), “A data warehouse is a subject-oriented, integrated, time-variant, nonvolatile collection of data in support of management’s decision making process.” What does Inmon mean by his four characteristics of a data warehouse? Subject-oriented means it focuses on subjects related to business or organizational activity like customers, employees and suppliers. Integrated means the data from various databases is stored in a consistent format through use of naming conventions, domain constraints, physical attributes, and measurements. Time-variant refers to associating data with specific points in time. Finally, nonvolatile means the data does not change once it is in the warehouse and stored for decision support. Ralph Kimball (1996), another data warehousing pioneer, states that “a data warehouse is a copy of transaction data specifically structured for query and analysis.” A related term is a “data mart.” A data mart is a more focused or a single- subject data warehouse. For example, some companies build a customer data mart rather than a multi-subject data warehouse. Such a focused data mart would have all of the business information about a company’s customers. Many 126 Decision Support Systems organizations and businesses are starting their enterprise-wide data warehouses by building a series of focused data marts. Data warehouses and data marts are often accessed using ad-hoc query or report and query tools. Some authors have combined data warehousing and OLAP. The two terms should be recognized as different subcategories of data-driven DSS. On-Line Analytical Processing (OLAP) OLAP and multidimensional analysis refers to software for manipulating multidimensional data. Even though one can have multidimensional data in a data warehouse, OLAP software can create various views and more dimensional representations of the data. According to Nigel Pendse at the OLAPReport.com, OLAP software provides fast, consistent, interactive access to shared, multidimensional information. Pendse calls these characteristics the FASMI test, an acronym for fast analysis of shared, multidimensional information test. What does the FASMI test mean? FAST means that the system delivers most responses to users within about five seconds. ANALYSIS means that the system can cope with any business logic and statistical analysis that is relevant for the application and the user. SHARED means that the software has security capabilities needed for sharing data among users. MULTIDIMENSIONAL is an essential requirement.